UCD9090: Logging?

Hi Khan

Please see my response below.

Regards

Yihe

Nabeel Khan said:

Part Number: UCD9090

Hi Team,

Can you please comment on the following issue?

- I set MISC_CONFIG bit0=1 to enable the brownout function.  I see that I have more than 5ms from 2.9V to 2.6V per the spec (I measured 24ms on my board).  However I don’t see logged_faults being saved across UCD power cycles. Is there other configuration needed?

[YH]no other setting is required. is the "store RAM to flash" button clicked to the save the settings? were there any faults presents during the test?did ucd log fault when brownout is disabled.

- With brownout enabled (MISC_CONFIG bit0=1) if I write all 0 to LOGGED_FAULTS (0xEA) entries, then power cycle the UCD, I see that LOGGED_FAULTS isn’t clear.  I think this issue could be related to the problem above.

[YH] was log fault cleared if brownout is disable? were there any faults after log was cleared?

- In UCD90xxx_PMBus_Command_Reference spec, section 9.26 it says "Note: The log must be cleared before the device begins normal operation.”  Does this  mean I should clear the fault log everytime the UCD powers up or only when it’s full? I did notice that if the log is full, even if I enable the FIFO logging (i.e. MISC_CONFIG bits 2 and 1) the log does not update. 

 [YH] customer need erase the log when it is full and logging FIFO mode is not eanbled if they want to log new faults. if the log is full, enable FIFO does not log new faults, this feature has to be enabled before log is full.

Here is a capture of V33A and V33D during power down:

Here is our UCD schematic for reference:

yihe said:

[YH]no other setting is required. is the "store RAM to flash" button clicked to the save the settings? were there any faults presents during the test?did ucd log fault when brownout is disabled.

I set MISC_CONFIG bit0=1 through I2C via our host CPU.  I was not using Fusion.  Once I set MISC_CONFIG bit0=1 via I2C do I need to store to flash and reset the UCD before the new configuration activates?  I did not do that, maybe that's the problem?

thanks
Jason

Hi Jason

We recommend this way.

Regards

Yihe

Hi Yihe,

 I can’t get MISC_CONFIG to save through power cycles. Is the following procedure correct?

set MISC_CONFIG to 0x7:  i2c block write to 0xfc with value 0x2, 0x7, 0x0

set STORE_DEFAULT_ALL = send byte 0x11

thanks
Jason

Hi Yihe,

I used TI Fusion GUI and probed the i2c bus while clicking the "Store RAM to Flash" button.  I saw that Fusion does a 1byte write to 0x11 with value 0x2a.  The bus transaction looks like:

<ucd_address><w><0x11><0x2a>

So I changed my host software to do the same and now the config is saved across power cycles.  Can you please tell me why the 0x2a is needed?

thanks!

jason

Hi Yihe,

Slave address is 7'b0110100 (0x34).  Do you mean the PEC 0x2a is optional? The Store to RAM definitely doesn't seem to work when I don't send 0x2a.

thanks
Jason

Hi Yihe,

I have the brownout function working but I'm having some issues clearing logged faults.  When I click the "Clear Logged Faults" button in Fusion, I see that all the log faults are clear.  However after I power cycle the UCD, the logged faults come back.  What's the proper procedure to clear the logged faults permanently? Do I need to do a "Store RAM to Flash" after clearing logged faults?

thanks
jason

Hi Yihe,

Thanks I see that the fault logs are clearing now.  My confusion  actually was due to the UCD logging numerous faults during power on which fills up the log. Is there a configuration that I can use to delay starting fault monitoring and logging after power on to allow the system to stabilize?  

Also what would be the recommendation in this situation:

- With logged faults not full, I enable FIFO and brownout features, save to flash, and reset the UCD

- During operation, logged faults becomes full

- If I power cycle the UCD, do I need to clear the logged faults before the FIFO function starts working again? Or should the FIFO function keep working across power cycles even if logged faults is full?

thanks
Jason